# train model CUDA_VISIBLE_DEVICES=0 \ python -u synthesis.py \ --max_len=300 \ --use_gpu=1 \ --output='./synthesis' \ --config='configs/ljspeech.yaml' \ --checkpoint_transformer='./checkpoint/transformer/step-120000' \ --checkpoint_vocoder='./checkpoint/vocoder/step-100000' \ if [ $? -ne 0 ]; then echo "Failed in training!" exit 1 fi exit 0